173

Catalysis Letters 48 (1997) 173^183

Kinetic study of the catalytic oxidation of alkanes over nickel,

palladium, and platinum foils
Mansour Aryafar and Francisco Zaera
Department of Chemistry, University of California, Riverside, CA 92521, USA
Received 12 August 1997; accepted 9 September 1997

The kinetics of the total oxidation of alkanes (methane, ethane, propane, n-butane and isobutane) over Ni, Pd, and Pt foils
was studied under fuel-lean conditions by using a recirculating batch reactor with mass spectrometry detection. Approximate firstand zeroth-order kinetics with respect to the hydrocarbon and oxygen concentrations, respectively, was observed in all cases.
Significantly slower rates were seen for methane conversion, but other subtle changes were also identified among the other hydrocarbons as well. The reactivity trends could be clearly correlated with C^H bond strengths, since higher rates were seen for the longer
and more branched hydrocarbon chains. Also, platinum was found to be the most active catalyst for the oxidation of all the compounds studied here except methane, which is oxidized faster on the nickel substrate. The variations in activity among the three catalysts were shown to be associated mostly with changes in the pre-exponential factor, not the activation energy, suggesting that they
have to do with the surface density of active sites on the surface. The nature of the active catalyst during reaction was determined by
simple inspection to be the metallic phase in the cases of Pt and Ni but an oxide layer in the case of Pd.
Keywords: total oxidation, alkanes, noble metals, kinetics, catalysis

1. Introduction
The catalytic total oxidation of hydrocarbons has
been investigated in the past in connection with a number of practical applications such as hydrocarbon detection in enclosed environments [1,2], odor control [3],
volatile organic compound removal from places such as
polymer processing, coating operations, spray painting
and offset printing [4,5], domestic [6,7] and other heating
systems [8,9], industrial boilers [10], gas turbines [11^
14], and fuel cells [15]. A renewed interest has surged in
this area recently in connection with hydrocarbon emission control from both automobiles and power plants
[16^19]. It was shown early on that noble metals, platinum and palladium in particular, are the most active for
promoting oxidation reactions, so a large effort has been
focused on studying catalysts based on those elements
[1,20^23]. Moreover, given that methane constitutes
more than 85% of the natural gas used in many modern
combustion engines and that it is the most difficult
alkane to burn, the majority of the work done over the
last decade has targeted the palladium/methane system
[24^39].
The kinetics and mechanism of the catalytic oxidation of hydrocarbons has indeed been studied extensively in the past. It has been found that the reaction
rates usually display approximately zeroth and first
order with respect to the partial pressures of oxygen
and the hydrocarbon, respectively, especially under the
fuel-lean conditions relevant to most pollution-control
* To whom correspondence should be addressed.
J.C. Baltzer AG, Science Publishers

applications [5,20,22,40,41]. This rate law has been

interpreted to be the consequence of an initial slow
activation step for the hydrocarbon on an oxygen-covered surface, a point of particular importance given
that the scission of the first C^H bond is still believed
to occur on a metallic or nearly metallic site [23]. In
fact, the nature of the active catalyst has been the subject of some controversy, but it is now widely agreed
that, while platinum remains mostly in its zero valence
state throughout the reaction, palladium owes its
unique reactivity to the formation and participation of
palladium oxide in the surface conversion steps; there
seems to be some synergism between palladium metal
and some form of PdOx [31,32,37,38,42,43]. In addition, hydrocarbon oxidation over noble metals is structure-sensitive [27,28,44,45], and often displays
induction periods [33,35], hysteresis [37,46], and aging
[28]. Different explanations have been provided for
this unique time-dependent behavior, but most argue
for possible changes in either morphology or oxidation
state of the palladium. Finally, the activity of the catalysts appears to be inhibited by the build-up of products (water and carbon dioxide) in the reaction
mixture [33,47] and by halides [26,35].
Unfortunately, in spite of the great practical importance of hydrocarbon oxidation reactions, most work
in this area has focused on a few specific systems, and
there has been very little systematic basic research in
terms of the active phase of the catalyst or the nature
of the reactants. A few trends have nevertheless been
reported already. For one, reasonably good correlations have been found between the ease with which the

174

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

alkanes are oxidized and the bond energy of their

weakest C^H bond [23,40^42,48^50]. It has been
clearly shown that the activation energy for the total
oxidation of paraffins decreases with chain size, but it
is not as clear if branching favors or hinders the reaction. The implication of this is, as mentioned above,
that C^H bond activation is most likely the rate-limiting step of the whole process; this conclusion has been
corroborated by isotopic substitution experiments, and
it is widely accepted. In terms of the catalyst, noble
metals have been proven much more active than oxides, but some contradictory reports can be found on
the interdependence between reactivity and surface
oxygen bond energy [20,22,51]. Also, platinum is
recognized to be the most active material for the oxidation of most alkanes, but a higher reactivity has been
reported repeatedly for palladium in the case of
methane [33,41,42,44,50]. Other late transition metals
(Rh, Ir, Ru) were found also to display high activities
towards hydrocarbon oxidations [22,41], but almost no
work has been published on nickel.
In this report we explore some of the points related
to the catalytic oxidation of alkanes using noble metals
enumerated above. The kinetics of the oxidation of the
different C1 to C4 alkanes were measured over nickel,
palladium, and platinum foils under fuel-lean conditions, and the activities of the solid samples were correlated with the pretreatments used for cleaning and
with the nature of the surface before, during, and after
reactions. It was found that the rate laws in all cases
are approximately first order in the hydrocarbon and
zeroth order in oxygen, the same as in other studies.
Both reaction rates and activation energies were found
to decrease with increasing chain size, but the effect of
branching was not as obvious. Platinum was found to
be more active than palladium in all cases, but, interestingly, nickel was shown to be more active than the
other two metals for methane combustion. These differences in reactivity were shown to be related mostly
with changes in pre-exponential factors, not activation
energies, which means that they are likely to originate
from changes in density of surface active sites in the
catalyst. Also, the most active phases were identified
by visual inspection to be the oxide for Pd and the
metal in the cases of Pt and Ni.
2. Experimental
The experiments reported here were carried out in a
recirculating batch reactor described in detail elsewhere [52]. Briefly, the stainless-steel loop and reactor,
which has a total volume of 200 cm3 , is evacuated with
a mechanical pump to a base pressure of about
5  103 Torr and then filled with the reacting gases
(oxygen and the hydrocarbon) and topped off to a final
pressure of 500 Torr by adding argon. The gases are

mixed by using a recirculation bellows pump, and the

time evolution of the partial pressures of the reactants
and products is followed with a quadrupole mass
spectrometer located in a separate chamber and connected to the reactor via a capillary tube. The collection and storage of data for further analysis is
accomplished by using a personal computer interfaced
to the mass spectrometer. Foils of about 8 by 12 mm2
in area were used as catalyst samples. They were spotwelded to two tantalum bars connected to copper feedthroughs in order to allow for resistive heating, and
their temperature was set by using an Eurotherm temperature controller and continuously monitored with a
chromel^alumel thermocouple spot-welded to their
back. Blank experiments were carried out both with
contaminated samples and with gold and tantalum
foils in order to ensure that the activity observed during the kinetic runs was entirely due to the noble metal
being tested; propane oxidation with a tantalum sample was seen to start above 1100 K, a temperature
approximately 450 K higher than when palladium was
used instead. Lastly, a viewport was also available in
the sample compartment for visual inspection.
The 15, 29, and 30 amu hydrocarbon peaks in their
mass spectra were used to follow the partial pressures of
methane, propane, and ethane, respectively, and the 58
amu signal was utilized for both n- and isobutane. The
18, 32, and 44 amu peaks were employed to follow the
time evolution of the pressures of water, oxygen, and
carbon dioxide, respectively, as long as there was no significant interference from the fragmentation of the
hydrocarbons. The mass spectrometer signal intensities
of the reacting gases were calibrated using a Baratron
pressure gauge and converted to partial pressures. Rates
and rate constants were then calculated afterwards by
using the ideal gas law and the area of the foils. Finally,
turnover numbers (TN) and turnover frequencies (TF)
are reported in molecules per metal surface atom and in
TN per second, respectively, and were estimated by
assuming an atomic surface density of approximately
1:5  1015 metal atoms/cm2 .
The metal foils were purchased from Aldrich
(99.9% minimum purity), and usually pretreated in situ
before each reaction. The Pt and Ni samples were
cleaned by successive exposures to oxygen and hydrogen at 1000 K, about 5 min each, and by then annealing to 1000 K under vacuum for a few minutes
afterwards, and the Pd foils were treated the same way
except that the reduction step was carried out with CO
instead of H2 to avoid palladium hydride formation.
Additional pretreatments were explored for the cases
of Pd and Ni, as discussed in more detail below. The
gases were all purchased from Matheson, and used as
supplied. All the hydrocarbons were of 99.5% minimum purity except for methane, which was 99.999%
pure, and the oxygen, hydrogen, and argon were
99.999% pure as well.

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

3. Results
3.1. General procedures
The kinetics of the catalytic oxidation of several saturated hydrocarbons over Ni, Pd, and Pt foils were studied under fuel-lean conditions. Hydrocarbon partial
pressure between 3 and 30 Torr and oxygen partial pressure between 60 and 300 Torr were used in most experiments. The reactivity of all three metal foils for the
oxidation of each of the hydrocarbons was first investigated by a temperature-programmed oxidation method
in which the extent of the conversion was followed as the
temperature of the catalyst was ramped at a rate of about
5 K/min. Figure 1 shows a typical survey of conversion
versus temperature for the case of the oxidation of 3.5
Torr of methane with 60 Torr of oxygen over a Pd foil.
These data were used to determine an approximate lightoff temperature (the inflexion point in the graph) and to
choose the optimum temperature range to be used in
subsequent isothermal kinetic studies. For instance, in
the example in figure 1 the data show that methane conversion starts around 1050 K, lights off around 1115 K,
and becomes too fast to be measured with our set-up by
1200 K; a temperature window between 1058 and 1178

Figure 1. Data from a typical temperature-programmed oxidation

(TPO) run for methane over a palladium foil under fuel-lean conditions.
The recirculation micro-batch reactor was filled sequentially with 3.5
Torr of methane, 60 Torr of oxygen, and 440 Torr of argon, and the
sample was then heated at a rate of 5 K/min while gas aliquots were
taken periodically via a capillary tube and analyzed by using a mass
quadrupole. This figure only displays the temporal evolution of the partial pressures of methane and carbon dioxide, but those of water and
oxygen were followed as well; no other compounds could be detected in
the gas phase. A light-off temperature of about 1115 K was measured in
this case. Similar experiments were performed for all the hydrocarbons
and all the metals employed in this study (see text).

175

K was therefore chosen for the isothermal measurements

in this case. These results also indicate that the extent of
methane conversion mirrors the amount of carbon dioxide formed, proving that the hydrocarbon undergoes
stoichiometric total oxidation. This was true for all cases
studied here (no other products were seen in the mass
spectra of the gases).
Isothermal kinetic measurements were performed
next. Figure 2 displays a representative isothermal catalytic run for the oxidation of methane over the Pd foil.
A mixture of 3.5 Torr of methane and 80 Torr of oxygen
was used in this case, and the reaction was carried out at
1178 K. This figure shows the evolution of the partial
pressures of methane, oxygen, and carbon dioxide as a
function of reaction time; the data for water was too
noisy to be useful because of the high background pressure of this compound in the mass spectrometer. Note
again that under the fuel-lean conditions used in these
experiments the hydrocarbon combustion is stoichiometric and complete: the only products detected in these
runs were carbon dioxide and water, and similar kinetic
data were obtained by following the partial pressures of
oxygen, carbon dioxide, and the hydrocarbon after
accounting for the stoichiometry of the reaction (which

Figure 2. Results from a typical isothermal kinetic run for the oxidation
of alkanes on metal foils using our experimental set-up. The case presented here is that of the conversion of 3.5 Torr of methane and 80 Torr
of oxygen over a palladium sample. The total pressure of the reactor
loop was taken up to about 500 Torr with argon in order to facilitate the
operation of the recirculation pump, and the solid was then heated to a
constant temperature of 1178 K by using a commercial controller.
Periodic gas samples were analyzed by mass spectrometry via a capillary tube, the same as in the TPO experiments. Shown in this figure is
the temporal evolution of the partial pressures of methane, oxygen, and
carbon dioxide; the pressure of water, the other product, was difficult
to follow because of the high background signal in the vacuum chamber
hosting the mass spectrometer.

176

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

in this case is CH4 2O2 ! CO2 2H2 O). Since the

data for the rate of disappearance of the hydrocarbon
usually had the least noise, those are the ones reported in
the remainder of this paper.
Reaction rate constants were calculated from data
such as those in figure 2 by either one of the two methods
illustrated in figure 3. In one (left panel of figure 3) initial
rates were estimated first from the slope in the pressure
versus time plots at early times (but after the first couple
of minutes, because there is an induction period due to
the delay induced by the slow diffusion of the gas sample
into the mass spectrometer), and rate constants k were
then calculated by dividing those numbers by the initial
hydrocarbon partial pressure. The second approach is
based on using first-order kinetics, by assuming that the
rate is proportional to the partial pressure of the hydrocarbon (see below). Such rate law makes semilogarithmic plots such as that in the right panel of figure 3, of
0
=PHC vs. time, yield a straight line with a slope
lnPHC
which corresponds directly to k. The two approaches
described in this paragraph yielded similar results for k.
3.2. Determination of kinetic parameters
The pressure dependence of the reaction rates for the
oxidation of each of the alkanes on all three metals was
determined first. An example of the results from those
studies is given in figure 4, which depicts data for the case
of n-butane oxidation over Ni foils. The changes in the
initial turnover frequency as a function of the partial
pressures of the alkane (left) and oxygen (right) were
plotted here in a log^log manner in order to extract reac-

tion orders (which are given directly by the slope of those

graphs). In most cases a constant oxygen pressure of 60
Torr was used in the first set of experiments as the pressure of the hydrocarbon was varied from 1 to 5 Torr, and
the alkane pressure was fixed at 3 Torr while the oxygen
pressure was changed from 20 to 100 Torr in the second.
In all cases it was found that the rate law for the oxidation is described at least approximately by first-order
kinetics with respect to the concentration of the hydrocarbon and zeroth-order dependence with respect to the
concentration of oxygen, in agreement with previous
reports [5,20,22,40,41]. These rate laws were also corroborated by the linearity of plots such as that in the right
panel of figure 3. Small differences were nevertheless
observed among the individual cases, specially as far as
the oxygen pressure dependence is concerned (table 1).
For instance, a small positive order, on the order of 0.2^
0.4, was measured for most hydrocarbons on nickel,
while on platinum and palladium the dependence was
often mildly negative. These differences may not be significant because they may be related to the temperature
at which the pressure-dependence kinetic studies were
carried out and that had to be chosen for each hydrocarbon^metal combination based on its specific reactivity
behavior; although all surfaces are expected to be almost
completely covered with oxygen under the fuel-lean conditions used here, the actual oxygen coverage may be
affected by the reaction temperature. This is a point
worth exploring in more detail, since it can shed more
light into the nature of the active site for the initial alkane
activation step [49,51].
Isothermal kinetic measurements were also carried

Figure 3. Illustration of the procedures followed to calculate reaction rate constants (k) from isothermal kinetic data such as those shown in
figure 2. Left: methane partial pressure versus time for the reaction between 3.5 Torr of methane and 80 Torr of oxygen over a palladium foil at
1178 K. The slope of the initial decay was divided by the initial partial pressure of the hydrocarbon to estimate the value of k. Right: semilogarithmic plot of the same data as in the left. Since the rate laws for most oxidation reactions are first order in the hydrocarbon and zeroth order in oxygen
(see figure 4 and text), these semilogarithmic plots yield (after a small induction period associated with the sample delay in our instrument) straight
lines, the slopes of which provide the value of k directly.

177

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

Figure 4. Pressure dependence studies for the oxidation of n-butane with oxygen over a nickel foil. Left: the pressure of the hydrocarbon was varied
from 1 to 5 Torr while that of oxygen was kept at 60 Torr to determine the approximately linear dependence between the reaction rate and the concentration of the n-butane. Right: the partial pressure of oxygen was varied between 20 and 100 Torr while keeping that of n-butane constant at 3
Torr in order to show the nearly independence of the rate on oxygen concentration in this pressure regime. All these experiments were carried out
at a total pressure of 500 Torr (by using argon as the ballast gas) and a foil temperature of 948 K. Experiments with other alkanes and metal foils
yielded similar results (table 1).

out as a function of temperature for all the alkanes and

all the metals studied here. The results obtained for the
rate constants are reported in Arrhenius plots in figure 5,
and the kinetic parameters, namely, the activation energies and pre-exponential factors calculated from those
data, are summarized in table 2. Notice that there is a
``rollover'' at the high temperature end of some of the
plots, in particular for the case of platinum, but even
though a similar behavior has been reported previously
[49], we believe that in our case this may be the result of a
limitation in the rates that can be measured with our
experimental set-up (because of mass-transfer limitations in the capillary to the mass spectrometer).
Otherwise the kinetic parameters reported here do in
general follow the expected trends in terms of the structure of the hydrocarbon, that is, the activation energies

decrease with chain length. The energy barrier is particularly high for methane, for which a value of about 30
kcal/mol was measured here on all three metals. This is
in agreement with most studies, although numbers
between 18 and 48 kcal/mol have been reported in the literature [1,20,33,44,45]. Good agreement was also found
between the data obtained here for the other hydrocarbons and previous work in spite of the fact that significant deviations are evident among the different reports
(table 3): there is a particularly good match between our
data and those of Hiam et al. [48] for platinum, and not
as good but still acceptable agreement with the results of
Cullis et al. [49] for palladium. In terms of the pre-exponential factors, they were seen here to decrease with
alkane size for nickel and palladium, but to go through a
maximum in the case of platinum; their values cover the

Table 1
Parameters for the rate laws in the total oxidation of light paraffins on nickel, palladium, and platinum foils under fuel-lean conditions. The rate
a
b
PO
. The reaction conditions used were as follows: for Ni and Pd, PHC was varied between 1
is expressed by the empirical relationship R kPHC
2
and 5 Torr while keeping PO2 at 60 Torr, then PO2 was varied between 20 and 100 Torr while keeping PHC at 3 Torr; for Pt the same conditions were
used for all hydrocarbons except methane and ethane, for which PHC was varied between 10 and 30 Torr while PO2 was kept at 300 Torr, and then
PO2 was varied between 100 and 300 Torr while keeping PHC at 10 Torr. The volume of the reactor was always taken to a total pressure of 500 Torr
by using argon as the ballast gas
Nickel

methane
ethane
propane
n-butane
isobutane

Palladium

Platinum

T (K)

T (K)

T (K)

928
973
973
948
943

0.9  0.1
0.9  0.1
0.8  0.1
1.0  0.2
0.7  0.3

0.2  0.1
0.3  0.1
0.3  0.1
0:2  0.2
0.4  0.3

1178
833
753
688
718

1.0  0.1
1.0  0.1
1.0  0.1
1.0  0.1
1.1  0.1

0:1  0.1
0:1  0.1
0.0  0.2
0.0  0.2
0.0  0.2

1073
973
873
673
673

1.1  0.1
1.0  0.1
1.2  0.1
1.2  0.1
1.3  0.1

0:1  0.1
0.0  0.1
0:2  0.1
0:2  0.1
0.2  0.1

178

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

Figure 5. Arrhenius plots for the oxidation reaction between alkanes (methane, ethane, propane, and n- and isobutane) and oxygen on all three
metals (nickel, palladium, and platinum) studied here. Each value of k, the reaction rate constant, was determined by isothermal experiments such
as those illustrated in figures 2 and 3. The activation energies and pre-exponential factors resulting from this analysis are summarized in table 2.

range between 1  1015 and 3  1010 per surface atom

and second. Comparison with other work in this case is
not easy because of the different ways the pre-exponential factor is calculated, but the similar ignition temperatures [42,48] or temperature ranges [41,49] reported
before are an indication of the similarity with our
kinetics.
The trends in reactivity are better highlighted by temperature-programmed oxidation experiments such as
those described in the previous section. The changes seen
for a given metal towards different hydrocarbons were
probed first; figure 6 displays the sequence obtained for
the Pd foil. It was found that on that metal methane oxidation starts around 1060 K, while the other compounds
(ethane, propane, n-butane, and isobutane) ignite at
temperatures around 700 K; on Pt the values are
approximately 900 K for methane and 500^570 K for the
others, and on Ni all the alkanes burn at slightly above
800 K. Again, methane is always much harder to burn
than the other paraffins, and the ease with which the

other compounds are converted increases with hydrocarbon chain size.

The variations in activity induced by changing metals
are exemplified in figure 7. For all hydrocarbons other
than methane the ease with which they are oxidized follows the sequence Pt > Pd > Ni, but, interestingly, in
the case of methane it was found that nickel is the most
effective catalyst for this reaction. It is important to
point out in connection with this switch that nickel
behaves in a unique manner because: (1) there is not
much difference in reactivity for that metal between
methane and the other alkanes; (2) the temperature-programmed oxidation curves obtained with Ni do not display the nice sigmoidal shapes seen for the other two
metals; and (3) the isothermal runs display a noticeable
deviation from first-order kinetics with time. This latter
observation was accompanied by the slow build-up of a
nickel oxide film on the surface (see below). We therefore
propose that the active phase in the nickel case is the
metal, and that its activity decreases with time as a con-

Table 2
Activation energies and pre-exponential factors for the total oxidation of light paraffins on nickel, palladium, and platinum foils under fuel-lean
conditions. The reaction conditions used were as follows: for Ni and Pd, PHC 3 Torr, PO2 60 Torr, and PAr 440 Torr; for Pt,
PCH4 ;C2 H6 30 Torr, PC3 H8 5 Torr, and PC4 H10 3 Torr, PO2 300 Torr for CH4 and C2 H6 and 60 Torr for C3 H8 and C4 H10 , PAr 170 Torr
for CH4 and C2 H6 , 440 Torr for C3 H8 and C4 H10
Nickel
T range
(K)
methane
ethane
propane
n-butane
isobutane

868973
868973
843948
803943
803943

Ea
(kcal mol1 )
28.1
27.8
26.5
18.2
18.0

Palladium
A
(s1 cm2 )
1300
2700
1500
20
20

T range
(K)
10581178
693833
648753
618688
648718

Ea
(kcal mol1 )
26.7
12.4
13.4
13.0
12.5

Platinum
A
(s1 cm2 )

T range
(K)

100
2
10
10
10

7821173
573973
523873
498673
498673

Ea
(kcal mol1 )
32.1
27.3
23.3
17.0
13.4

A
(s1 cm2 )
6600
54000
540000
4900
150

179

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

Table 3
Comparison between the activation energies obtained here for the oxidation of hydrocarbons on noble metals and those reported previously.
All values are given in kcal/mol
Reference

Catalyst

CH4

C2 H 6

C3 H 8

n-C4 H10

iso-C4 H10

this work
[49]
[41]
[50]

palladium
foil
on Al2 O3 beads
wire
black

26.7
13.4
17
14.4

12.4
20.1
22

13.4
15.6
23

13.0

26
23.9

12.5
11.0

this work
[48]
[41]
[50]

platinum
foil
filament
wire
black

32.1

21
28.2

27.3
27.3
26

23.3
17.0
22

17.0
17.0
25
19.1

13.4
10.2

sequence of site blocking by the growing NiO coating.

It could be thought that steam reforming may play a role
in the nickel case, but this is unlikely because no CO was
detected by mass spectrometry at any point during the
kinetic runs.
3.3. Effect of sample pretreatment on the reaction
kinetics
Complementary experiments were performed by
varying the pretreatment used to clean the metal foils

Figure 6. Temperature-programmed oxidation data for the different

alkanes on a palladium foil. All these experiments were carried out by
using 3.5 Torr of the hydrocarbon, 60 Torr of oxygen, and 440 Torr of
argon, and by heating the foil at a rate of 5 K/min. A clear trend is seen
in this figure in terms of reactivity, as indicated by the shifts down in
light-off temperature seen with increasing hydrocarbon chain size. This
represents a direct correlation between the ease of oxidation of the paraffin and the bond energy of its weakest C^H bond.

before reactions in order to identify the active phase of

the catalyst. Four different pretreatment sequences were
tested here: (1) a quick flash of the surface to 1000 K in
vacuum before reactions; (2) exposure of the foil to 500
Torr of oxygen at 1000 K for about 5 min prior to flashing to the same temperature under vacuum; (3) sequential 5 min treatments with 500 Torr of oxygen and 500
Torr of hydrogen, both at 1000 K, and then pumping of
the system; and (4) sequential 5 min 500 Torr treatments
with oxygen and carbon monoxide at 1000 K followed
by pumping.
An indication of the nature of the catalyst both after
the different pretreatments and all throughout the
kinetic reaction runs was provided by the physical
appearance of their surfaces. In the case of palladium the
initial foil was shiny and of a metallic silvery color in
spite of the likelihood that the surface was covered by a
thin layer of contaminant hydrocarbons. Upon oxidation, however, that substrate turned greenish-blue, indicating the formation of palladium monoxide (PdO) [53].
After exposure to hydrogen the foil then appeared to
swell as it formed the palladium hydride (Pd2 H), and it
turned silvery gray (dull instead of shiny), became quite
brittle, and never recuperated its shine again. Because of
the significant physical changes induced by this latter
treatment, the foils exposed to hydrogen were not used
for any measurements other than those discussed in this
section. Lastly, reduction of the oxide with CO instead
of hydrogen did reverse the sample to its shiny metallic
state.
The behavior of the palladium foil during the formation of water induced by the oxygen hydrogen treatments was investigated in a bit more detail. When
hydrogen was added to an oxygen-pretreated palladium
foil, its color was seen to change from greenish-blue
(PdO) to silvery gray (as mentioned above), the temperature was seen to increase by about 60^70 K without any
external heating, and large amounts of water were
detected by the mass spectrometer. When oxygen was
added to a hydrogen-pretreated palladium, on the other

180

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

Figure 7. Temperature-programmed oxidation data for methane (left) and isobutane (right) on all three (Ni, Pd, and Pt) foils. All the experiments
were carried out by using 3.5 Torr of the hydrocarbon, 60 Torr of oxygen, and 440 Torr of argon, and by heating the foil at a rate of 5 K/min.
Platinum proved to be the most active catalyst for the oxidation of all the hydrocarbons used here other than methane, but methane was shown to
burn at lower temperatures on the nickel foil.

hand, the color of the surface changed from silvery gray

(Pd2 H) back to greenish-blue (PdO), but significantly
less water was produced, and the temperature rise of the
foil was an order of magnitude larger (about 600^700
K). The difference in the temperature jumps seen in both
cases is most likely explained by the differences in the
amount of the heat released during PdO formation,
which is about 20:4 kcal/mol [53], as compared to that
from Pd2 H formation, which amounts to only 8:9 kcal/
mol [53].
The catalytic activity towards hydrocarbon total oxidation of the pretreated Pd foils was then probed at 753
K by using 3 Torr of propane and 60 Torr of oxygen.
Also, several runs were performed for each type of pretreatment in order to evaluate the reproducibility of the
treatments. This was needed because it was found that it
usually took between 5 and 15 cycles for the sample to
behave in a predictable manner. Figure 8 displays some
of the key results obtained for the propane oxidation rate
constants on Pd (and Ni) foils after the different pretreatment procedures: the hatched bars provide the
lower limit for the value of k, and the white bars on top
represent the spread obtained in the reproducibility
experiments. It can be seen in this figure that for the case
of palladium the oxygen CO pretreatment is clearly
the most effective in terms of both activity and reproducibility. It was also seen that the palladium hydride does
not remain in that state but converts into the (non-crystalline) metal during the early stages of the reaction, and
that the resulting sample can still catalyze hydrocarbon
oxidations with the same efficiency as a new foil. Indeed,
when the reaction is started by heating the hydride to 753
K in the presence of the reaction mixture, the color of the

foil changes almost immediately from silvery gray

(Pd2 H) to greenish-blue (PdO), the same as when exposing it to oxygen alone. The greenish-blue sample color
was in fact seen during the course of the reactions in all
cases, suggesting that palladium monoxide is the active
species which catalyzes the hydrocarbon oxidation
[20,33,49,50]. Interestingly, at the end of the reaction,
after all the hydrocarbon molecules were oxidized, the
foil heating was turned off to allow it to cool down, and
the gases were pumped out, the color of the foil became a
dull red indicative of palladium dioxide (PdO2
xH2 O)
formation.
Additional reproducibility experiments were
designed to check for self-poisoning during the oxidation
of propane over the Pd foil. The initial conditions were
kept identical to those used before, namely, the reactions
were started at 753 K by using a mixture of 3 Torr of propane and 60 Torr of oxygen, but the reactor was then
evacuated after a preset reaction time (about 5, 10, 15,
20, and 25 min) while keeping everything else unchanged
(and maintaining the sample at 753 K), and then refilled
with new reactants soon after. In these kinetic experiments the reaction was seen to pick up right from the
start at the same rate as the initial run, and 100% conversion of the second mixture was obtained in the expected
time interval. This indicates that the operating surface of
the catalyst is not modified by the reaction mixture
under the conditions used here.
Similar kinetic experiments following the different
pretreatment procedures were performed for the oxidation of propane on nickel at 873 K (figure 8). In the case
of the original Ni sample, after only flashing to 1000 K,
the foil eventually changed in color during the reaction,

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

181

tion runs; (2) the reaction rates slowed down with time
and deviated significantly from first-order behavior; and
(3) the kinetic data obtained here was not as reproducible as in the other cases (notice the size of the white bar
over the O2 H2 -pretreated Ni sample in figure 8).
Apparently it is not easy for nickel to retain its metallic
state under fuel-lean oxidation conditions.
Finally, the Pt samples were cleaned sequentially with
oxygen and hydrogen as is commonly done with supported catalysts [45,54,55], and the foil remained shiny
and metallic in all cases. No reproducibility problems
were encountered with this metal.
4. Discussion

Figure 8. Determination of k (the reaction rate constants) from isothermal kinetic runs designed to probe the effect of different pretreatments
on the performance of palladium and nickel foils for the total oxidation
of saturated hydrocarbons. All these experiments were carried out with
3 Torr of propane, 60 Torr of oxygen, and 440 Torr of argon at 753 K.
Results from four types of pretreatments are reported here, namely
(from left to right), (1) none; the catalyst was used immediately after a
quick flash to 1000 K under vacuum; (2) a 5 min pre-exposure to 500
Torr of oxygen at 1000 K; (3) 5 min sequential pre-exposures to 500
Torr of oxygen and hydrogen at 1000 K; and (4) 5 min sequential
pre-exposures to 500 Torr of oxygen and carbon monoxide at 1000 K
(done for the case of palladium only). The hatched bars represent the
lower-limit values of k obtained in each case, while the white bars on
top indicate the variations seen over a number of runs and illustrate the
stability and reproducibility of the different states of the catalysts. The
initially reduced metal foils appear to always perform better, and in the
case of palladium reduction with CO yields a more reproducible
surface.

from its shiny silvery metallic appearance (Ni) to greenblack (NiO
H2 O), but the activity was low: the measured
reaction rate constant was about 4  104 s1 cm2 . A
similar reactivity was seen when the catalyst was pretreated by heating it to 1000 K under 500 Torr oxygen,
even though in that case the foil was green-black from
the start. After cleaning the sample by heating to 1000 K
first under 500 Torr of oxygen and then in 500 Torr of
hydrogen, however, the measured average rate constant
was about 1:2  103 s1 cm2 , about three times larger
than in the other two cases. This change is easy to understand in view of the fact that the color of the oxygen-pretreated foil changed from green-blue (NiO) back to
shiny silvery (Ni) upon hydrogen exposures, proving
that the combined pretreatment with oxygen and hydrogen regenerates the metallic Ni surface, and that that
phase is the one active for the combustion reactions.
Three more observations are relevant to the catalytic
behavior of the nickel foil: (1) the metallic Ni surface was
seen to evolve slowly into the oxide during all the reac-

The data reported here broadens the existing database

for alkane total oxidation reactions on noble metal catalysts. To begin with, the results obtained in this study
corroborate some of the trends presented in previous
publications. In particular, a good correlation was
observed between C^H bond energies and reactivities,
suggesting that the initial activation of the alkane is the
rate-limiting step for the overall combustion process
[23]. Indeed, the activation energy for alkane oxidation
was seen to decrease with hydrocarbon chain length on
the three metals studied here [22,42,48^50]. Chain
branching seems to also facilitate alkane activation
somewhat (compare the rate constant values for n- and
isobutane in figure 5, and those of the activation energies
in table 1), but this effect is not as dramatic as in the case
of chain size [22,48,49,56], and the opposite trend has in
fact been reported on occasion [57]. Next, the conversion
of methane was also proven to be significantly harder
than that of any other hydrocarbon [22,23,58]: light-off
temperatures as much as 300 K higher than those of the
other alkanes were observed for that molecule in our
temperature-programmed oxidation experiments (see
for instance the data in figure 6).
New information was derived from this investigation
as well. In terms of the changes in oxidation activity
induced by changing the metal used as the catalyst, platinum foils proved to be more active than palladium
samples for all the alkanes studied here. There is little
discussion about the fact that platinum is more effective
than palladium in catalyzing the combustion of heavier
alkanes [22,42,50,59], but a higher activity has been
often reported for the oxidation of methane on palladium [1,20,44,50] unless reducing conditions are used
[29,60]. To explain the differences in activity trends
between our data and those in previous reports, it is
important to remember that the unique behavior of palladium-based catalysts towards methane conversion
has been ascribed to the presence of some type of oxide
on the surface [20,33,49,50]. Here it was also shown
that, while the active phase in the case of platinum was
the metal, it is indeed the oxide that promotes oxidation

182

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

with palladium. However, the actual nature of the

active site on the palladium-based catalysts has been
debated to a great extent in the literature, because it is
believed that there is some synergism between the metal
and some form of oxygen-containing surface species in
a way that makes both oxidation states indispensable
for the activity of the catalyst [31,32,37,38,42,43]. In
that regard we point to the fact that the main differences in activity reported here are manifested as variations in pre-exponential factors; the activation energy
for methane oxidation was in fact lower on palladium
than on platinum. We would argue that these variations
in the pre-exponential term are likely to reflect changes
in the density of active sites. We propose that the low
values of A measured here for Pd (table 2) reflects the
need of unique sites which require a particular combination of metal and oxide components [38] present only in
low concentrations on the surface of the foil. It is also
possible that the high values obtained for platinum
reflect the ease with which that metal can be kept in its
metallic state.
There is another conclusion that can be derived from
the fact that the values of the pre-exponential factors
for alkane oxidation reactions measured here are quite
low. While transition state theory (and experimental
results for other surface decomposition reactions) predict numbers of the order of 1017 ^1020 s1 cm2 [61], the
largest pre-exponential factor in table 2 is that of
propane on Pt, and that reached a value of only about
5  105 s1 cm2 . This difference can be easily explained by remembering that besides a collision frequency term, the A factor as calculated here includes
the effect of the sticking probability of the reactant,
which is particularly low for alkanes on metals [62^66].
In that respect, our observation provides additional
support for the idea of the hydrocarbon activation
being the rate-limiting step for the overall oxidation. It
is also interesting to notice that, while in the case of palladium the pre-exponential factor decreases pretty
much monotonically with increasing alkane chain size,
for platinum (and to a lesser extent nickel) this parameter goes through a maximum between ethane and propane. Although we do not have an explanation for this
behavior at the present time, it appears to account for
the reversing in the order of activity between platinum
and palladium when going from methane to heavier
hydrocarbons [60].
One final point worth bringing up here is the fact
that there is very little work published on the catalytic
behavior of nickel-based catalysts for oxidation reactions. NiO has been shown to be one of the most active
oxides in the catalysis of alkane oxidation [22,51], but
the reaction rates on that substrate are still several
orders of magnitude lower than those seen on platinum
or palladium under the same conditions [1,59]. We are
not aware of any previously published work on the
reactivity of metallic nickel for hydrocarbon total com-

bustion. Supported nickel catalysts have been shown

to be quite efficient for syngas formation as long as the
contact times are kept low [67,68], but no evidence for
such a pathway was obtained in our batch-reactor
experiments. Instead, it was found here that methane
total combustion is significantly faster on nickel than
on either palladium or platinum (figures 5 and 7). It
needs to be pointed out, however, that we found that
metallic nickel is the active phase for this reaction, and
that such oxidation state is difficult to maintain under
the fuel-lean conditions used in these experiments: a
slow-down of the reaction rate beyond that expected
from first-order kinetics was seen during each individual kinetic run, and the film was seen to become oxidized over time. Nevertheless, it is somewhat puzzling
that no significant differences were observed between
the kinetics of oxidation of methane and that of the
other alkanes on the nickel foil, as it was seen on the
other two metals. We believe that in spite of the potential problems related with maintaining the proper oxidation state of the catalyst during reactions, it is
worthwhile looking into nickel-based catalysts as alternatives for the design of alkane emission control
systems.
5. Conclusions
Kinetic data were obtained in this investigation for
the total oxidation of methane, ethane, propane, nbutane, and isobutane on nickel, palladium, and platinum foils under fuel-lean conditions by using a microbatch recirculating reactor. Clear trends were seen on
each metal as a function of the nature of the hydrocarbon, the activity increasing with chain size and to
a lesser extent with chain branching. The rate laws
were in all cases approximately first and zeroth order
with respect to the hydrocarbon and oxygen partial
pressures, respectively, but some differences were seen
in individual cases, presumably because of the different reaction temperatures used in the pressure-dependence determinations. The reactivity of the three
metals towards the oxidation of each of the alkanes
studied here was contrasted by performing both temperature-programmed oxidation and isothermal oxidation experiments. The activity for the conversion of
all the paraffins other than methane was found to follow the trend Pt > Pd > Ni, but methane oxidation
was proven to be faster on the nickel foil. By observing the color of the foils during reactions it was
determined that the active catalytic phases in the
cases of Pt and Ni are their metallic form, but for Pd
it is the oxide instead. Most of the differences in reactivity seen among the metals were ascribed to changes
in pre-exponential factors, an indication of changes in
the surface density of active sites with changing
catalyst.

M. Aryafar and F. Zaera / Kinetics of catalytic alkane oxidation

Acknowledgement
Financial support for this research was supplied
mainly by a grant from the National Science Foundation
(CTS-9525761). Additional funds were provided by a
seed grant from the University of California Energy
Institute.
References
[1] R.B. Anderson, K.C. Stein, J.J. Feenan and L.J.E. Hofer, Ind.
Eng. Chem. 53 (1961) 809.
[2] J.G. Firth, Trans. Faraday Soc. 52 (1966) 2566.
[3] R.E. Roberts and J.B. Roberts, J. Air Pollut. Control Assoc. 26
(1976) 353.
[4] A.E.R. Budd, Environ. Pollut. Manag. (1978) 13.
[5] J.J. Spivey, Ind. Eng. Chem. Res. 26 (1987) 2165.
[6] M.R. Dongworth and A. Melvin, 16th Int. Symp. in
Combustion, MIT, 1976, p. 255.
[7] A. Nishino, Catal. Today 10 (1991) 107.
[8] S.W. Radcliffe and R.G. Hickman, J. Inst. Fuel 48 (1975) 208.
[9] D.L. Trimm and C.W. Lam, Chem. Eng. Sci. 35 (1980) 1731.
[10] W.V. Krill, J.P. Kesselring, E.K. Chu and R.M. Kendall, Mech.
Eng. 102 (1980) 28.
[11] S.M. DeCorso, S. Mumford, R.V. Carruba and R. Heck, J. Eng.
Power 99 (1977) 159.
[12] W.C. Pfefferle, J. Energy 2 (1978) 142.
[13] B.E. Enga and D.T. Thompson, Plat. Met. Rev. 23 (1979) 134.
[14] J. Tsuji and K. Ohno, Synthesis 157 (1969) 1.
[15] J.P. Kesserling, in: Advanced Combustion Methods, ed.
F.J. Weinberg (Academic Press, New York, 1986) p. 237.
[16] J. Wei, Adv. Catal. 24 (1975) 57.
[17] K.C. Taylor, Automobile Catalytic Converters (Springer, Berlin,
1984).
[18] R.J. Farrauto, R.M. Heck and B.K. Speronello, Chem. Eng.
News (7 Sept.) (1992) 34.
[19] H.S. Gandhi and M. Shelef, in: Catalysis and Automotive
Pollution Control, eds. A. Crucq and A. Frennet (Elsevier,
Amsterdam, 1987) p. 199.
[20] J.G. Firth and H.B. Holland, Trans. Faraday Soc. 65 (1969)
1121.
[21] D.L. Trimm, Appl. Catal. 7 (1983) 249.
[22] G.I. Golodets, Heterogeneous Catalytic Reactions Involving
Molecular Oxygen, Studies in Surface Science and Catalysis
Series, Vol. 15 (Elsevier, Amsterdam, 1983).
[23] R. Burch and M.J. Hayes, J. Mol. Catal. A 100 (1995) 13.
[24] M. Niwa, K. Awano and Y. Murakami, Appl. Catal. 7 (1983)
317.
[25] C.F. Cullis and B.M. Williatt, J. Catal. 86 (1984) 187.
[26] R.F. Hicks, H. Qi, M.L. Young and R.G. Lee, J. Catal. 122
(1990) 295.
[27] T.R. Baldwin and R. Burch, Appl. Catal. 66 (1990) 359.
[28] P. Briot and M. Primet, Appl. Catal. 68 (1991) 301.
[29] S. Oh, P.J. Mitchell and R.M. Siewert, J. Catal. 132 (1991) 287.
[30] K. Otto, L.P. Haack and J.E. De Vries, Appl. Catal. B 1 (1992)
1.
[31] R.J. Farrauto, M.C. Hobson, T. Kennelly and E.M. Waterman,
Appl. Catal. A 81 (1992) 227.
[32] P.E. Marti, M. Maciejewski and A. Baiker, J. Catal. 141 (1993)
494.

183

[33] F.H. Ribeiro, M. Chow and R.A. Dalla Betta, J. Catal. 146
(1994) 537.
[34] E. Garbowski, C. Feumi-Jantou, N. Mouaddib and M. Primet,
Appl. Catal. 109 (1994) 277.
[35] S.S. Peri and C.R.F. Lund, J. Catal. 152 (1995) 410.
[36] R. Burch and F.J. Urbano, Appl. Catal. A 124 (1995) 121.
[37] P. Salomonsson, S. Johansson and B. Kasemo, Catal. Lett. 33
(1995) 1.
[38] S. Su, J. Carstens and A.T. Bell, 15th Meeting North American
Catalysis Society, Chicago 1997, paper D21, p. 91.
[39] K. Fujimoto, F.H. Ribeiro, M. Avalos-Borja and E. Iglesias,
15th Meeting North American Catalysis Society, Chicago 1997,
paper D23, p. 93.
[40] M.A. Accomazzo and K. Nobe, Ind. Eng. Chem. Proc. Res.
Develop. 4 (1965) 425.
[41] Y.-F.Y. Yao, Ind. Eng. Chem. Prod. Res. Develop. 19 (1980)
293.
[42] A. Schwartz, L.L. Holbrook and H. Wise, J. Catal. 21 (1971)
199.
[43] C.A. M
uller, M. Maciejewski, R.A. Koeppel and A. Baiker, J.
Catal. 166 (1997) 36.
[44] R.F. Hicks, H. Qi, M.L. Young and R.G. Lee, J. Catal. 122
(1990) 280.
[45] K. Otto, Langmuir 5 (1989) 1364.
[46] J.G. McCarty, Catal. Today 26 (1995) 283.
[47] R. Burch, F.J. Urbano and P.K. Loader, Appl. Catal. A 123
(1995) 173.
[48] L. Hiam, H. Wise and S. Chaikin, J. Catal. 9/10 (1968) 272.
[49] C.F. Cullis and T.G. Nevell, Proc. Roy. Soc. London A 349
(1976) 523.
[50] V.A. Drozdov, P.G. Tsyrulnikov, V.V. Popovskii,
N.N. Bulgakov, E.M. Moroz and T.G. Galeev, React. Kinet.
Catal. Lett. 27 (1985) 425.
[51] M.F.M. Zwinkels, S.G. J
aras and P.G. Menon, Catal. Rev.-Sci.
Eng. 35 (1993) 319.
[52] A. Loaiza, M. Xu and F. Zaera, J. Catal. 159 (1996) 127.
[53] R.C. Weast, ed., CRC Handbook of Chemistry and Physics
(CRC Press, Cleveland, 1974).
[54] G.C. Bond, Catalysis by Metals (Academic Press, London,
1962).
[55] B.C. Gates, J.R. Katzer and G.C.A. Schuit, Chemistry of
Catalytic Processes (McGraw-Hill, New York, 1979).
[56] M.F. Umstead, F.G. Woods and J.E. Johnson, J. Catal. 5 (1966)
293.
[57] J.E. Johnson, J.G. Christian and H.W. Carhart, Ind. Eng.
Chem. 53 (1961) 900.
[58] J.R. Regalbuto, An Analysis of Post-Combustion Catalytic
Emissions Treatment, Gas Research Institute, Final Task Report
GRI 92/0445 (1992).
[59] Y. Moro-oka, Y. Morikawa and A. Ozaki, J. Catal. 7 (1967) 23.
[60] J. Tsuji and K. Ohno, Tetrahedron Lett. (1970) 823.
[61] K.J. Laidler, Chemical Kinetics, 3rd Ed. (Harper and Row, New
York, 1987).
[62] C.T. Rettner, H.E. Pfn
ur and D.J. Auerbach, Phys. Rev. Lett.
54 (1985) 2716.
[63] S.G. Brass and G. Ehrlich, J. Chem. Phys. 87 (1987) 4285.
[64] A.C. Luntz and D.S. Bethune, J. Chem. Phys. 90 (1989) 1274.
[65] M.C. McMaster and R.J. Madix, Surf. Sci. 275 (1992) 265.
[66] W.H. Weinberg, Langmuir 9 (1993) 655.
[67] S.S. Bharadwaj and L.D. Schmidt, J. Catal. 146 (1994) 11.
[68] M. Huff, P.M. Torniainen and L.D. Schmidt, Catal. Today 21
(1994) 113.